Compaction freezes

[Michał Łowicki <ml...@gmail.com>]

Hi,

Using C* 2.1.5 and table with leveled compaction I've found that number of
pending tasks is around 100. Turned out that one table cannot be compacted
and it always stops at the same point (more or less):

Compaction       sync   entity_by_id     4.91 GB   12.34 GB   bytes
39.79%

`service cassandra restart`

Compaction       sync   entity_by_id     4.91 GB   12.35 GB   bytes
39.74%

`service cassandra restart`

Compaction       sync   entity_by_id      4.9 GB   12.33 GB   bytes
39.77%

`service cassandra restart`

Compaction       sync   entity_by_id     4.89 GB   12.32 GB   bytes
39.73%

After doubling heap size (cassandra-env.sh) compaction went fine but I'm
considering this change as a temporary solution and after a while
compaction started to freeze again anyway.

Is there any way to get insight into compaction so to get answer why it
freezes and basically what compactor is doing currently? I've enabled DEBUG
logging but it's too verbose as node is getting some traffic. Can I enable
DEBUG for compaction only?
-- 
BR,
Michał Łowicki